Hindawi Publishing Corporation Neural Plasticity Volume 2016, Article ID 4680972, 12 pages http://dx.doi.org/10.1155/2016/4680972

Research Article Abnormal Resting-State Functional Connectivity Strength in Mild Cognitive Impairment and Its Conversion to Alzheimer’s Disease Yuxia Li,1,2 Xiaoni Wang,1 Yongqiu Li,2 Yu Sun,1 Can Sheng,1 Hongyan Li,1 Xuanyu Li,1 Yang Yu,1 Guanqun Chen,1 Xiaochen Hu,3 Bin Jing,4 Defeng Wang,5 Kuncheng Li,6 Frank Jessen,3 Mingrui Xia,7 and Ying Han1,8 1

Department of Neurology, XuanWu Hospital of Capital Medical University, Beijing 100053, China Department of Neurology, Tangshan Gongren Hospital, Tangshan 063000, China 3 Department of Psychiatry and Psychotherapy, University Hospital of Cologne, Kerpener Strasse 62, 50937 Cologne, Germany 4 School of Biomedical Engineering, Capital Medical University, Beijing 100069, China 5 Research Center for Medical Image Computing, Department of Imaging and Interventional Radiology, The Chinese University of Hong Kong, Shatin, Newterritories 000852, Hong Kong 6 Department of Radiology, XuanWu Hospital of Capital Medical University, Beijing 100053, China 7 State Key Laboratory of Cognitive Neuroscience and Learning and IDG/McGovern Institute for Brain Research, Beijing Normal University, Beijing 100875, China 8 Center of Alzheimer’s Disease, Beijing Institute for Brain Disorders, Beijing 100053, China 2

Correspondence should be addressed to Mingrui Xia; [email protected] and Ying Han; [email protected] Received 5 August 2015; Accepted 4 October 2015 Academic Editor: Feng Shi Copyright © 2016 Yuxia Li et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Individuals diagnosed with mild cognitive impairment (MCI) are at high risk of transition to Alzheimer’s disease (AD). However, little is known about functional characteristics of the conversion from MCI to AD. Resting-state functional magnetic resonance imaging was performed in 25 AD patients, 31 MCI patients, and 42 well-matched normal controls at baseline. Twenty-one of the 31 MCI patients converted to AD at approximately 24 months of follow-up. Functional connectivity strength (FCS) and seedbased functional connectivity analyses were used to assess the functional differences among the groups. Compared to controls, subjects with MCI and AD showed decreased FCS in the default-mode network and the occipital cortex. Importantly, the FCS of the left angular gyrus and middle occipital gyrus was significantly lower in MCI-converters as compared with MCI-nonconverters. Significantly decreased functional connectivity was found in MCI-converters compared to nonconverters between the left angular gyrus and bilateral inferior parietal lobules, dorsolateral prefrontal and lateral temporal cortices, and the left middle occipital gyrus and right middle occipital gyri. We demonstrated gradual but progressive functional changes during a median 2-year interval in patients converting from MCI to AD, which might serve as early indicators for the dysfunction and progression in the early stage of AD.

1. Introduction Alzheimer’s disease (AD), an irreversible neurodegenerative disease characterized by memory dysfunction, executive function decline, and multiple cognitive domain impairments, is one of the most financially costly diseases [1]. Since there is currently no effective treatment to stop or reverse the progression of AD, the research spotlight has turned to its

predementia stage, specifically termed amnestic mild cognitive impairment (aMCI). For individuals with MCI due to AD (called “aMCI” or MCI in this paper for short), the development of AD is a high risk factor that the rate of MCI conversion to AD reaches 10% to 15% annually [2]. Considering the urgent requirement for the identification of those MCI patients who are most likely to undergo rapid progression and conversion to AD, it is of great significance to investigate and

2 discover the potential biomarkers for the early identification of the dysfunction and progression in the early stage of AD. Magnetic resonance imaging, a noninvasive, nonradiation means for the mapping of both structures and functions of the human brain, is a promising avenue to investigate the progressive brain changes from MCI to AD [3, 4]. Structurally, studies have consistently found that the gray matter atrophy originally starts at the medial temporal lobe, spreads along the midline of the cerebral cortex, and finally extents to the whole brain during the progress from MCI to AD [5– 7]. Functionally, however, investigations have yielded limited functional biomarkers that predict the progression from MCI to AD, except the consistent identification of the changes of resting-state functional connectivity (RSFC) of the defaultmode network (DMN) in AD [8, 9]. However, deficits in RSFC are not confined to the DMN in patients with MCI converting to AD [10]. Furthermore, it is not clear whether other brain regions participate in the conversion to AD. Most previous studies have focused on the AD- or MCIrelated functional connectivity changes of specific predefined regions of interest, such as posterior cingulate cortex and thalamus [11, 12]. Given the complex pathology and widespread functional abnormalities in AD and MCI, it would be of great interest to examine differences between MCI-converters (MCI-c) and MCI-nonconverters (MCI-nc) within a wholebrain range. Here, we used resting-state functional magnetic resonance imaging (R-fMRI) data and functional connectivity strength (FCS), computed as the sum of connections between a given voxel and all other voxels [13–15], to detect the functional differences among AD, MCI, and normal controls and especially between MCI patients who converted to AD (MCI-c) and MCI-nc. We sought to determine (1) whether there exists an AD-related progressive abnormality pattern on the whole-brain functional connectivity strength in MCI patients and (2) if so whether these changes are different between MCI-c and MCI-nc groups and are related to their clinical behaviors.

2. Materials and Methods 2.1. Participants. The study was approved by the Research Ethics Review Board of XuanWu Hospital (ClinicalTrials.gov Identifier: NCT02353845). A total of 98 right-handed subjects were recruited in the study including 25 AD patients, 31 MCI patients, and 42 well-matched cognitive normal controls. All AD and MCI patients were recruited at the memory clinic of the Neurology Department, XuanWu Hospital, Capital Medical University, Beijing, China. Control subjects were recruited from the local community via broadcast and advertisements. Diagnoses of MCI due to AD were made by experienced neurologists using Petersen’s criteria [16]. The diagnosis of AD fulfilled the published diagnostic criteria [17]. Controls were screened as described in the Structured Interview for DSMIV Nonpatient Edition [18] to confirm the life-long absence of psychiatric and neurological illness. Inclusion criteria for MCI due to AD included the following: (1) memory complaint, preferably confirmed by an informant; (2) objective memory impairment, (cutoff points of Mini-Mental State Examination (MMSE) score [19]: 19 (no formal education),

Neural Plasticity 22 (1 to 6 years of education), and 26 (7 or more years of education); cutoff points of Montreal Cognitive Assessment (MoCA) [20]: 13 (no formal education), 19 (1 to 6 years of education), and 24 (7 or more years of education); cutoff point of auditory verbal learning test- (AVLT-) delayed recall [21]: 6); (3) no or minimal impairment of daily life activities; (4) a Clinical Dementia Rating (CDR) [22] score of 0.5; (5) being free from dementia according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, revised (DSMIV-R) [18]; (6) hippocampal atrophy confirmed by structural MRI; and (7) the Han nationality, right-handed (the Edinburgh handedness scale score [23] >40 points). The exclusion criteria applied to all subjects with contraindications for MRI; also excluded were those with histories of stroke, psychiatric disease, neurological disorder, alcohol or drug abuse, and systemic disease such as severe anemia, thyroid dysfunction, syphilis, or Acquired Immune Deficiency Syndrome. All subjects underwent a standardized clinical and neuropsychological evaluation, including the MMSE, MoCA, clock drawing test (CDT), AVLT, activities of daily living scale, Hachinski Ischemic Scaling, Hamilton Depression Scale, and CDR. Second, the quality of the whole-brain resting-state functional MRI images was inspected by an experienced neuroradiologist. Third, after a mean follow-up period of 24 months (ranging from 11 months to 48 months), subjects again underwent the entire clinical and neuropsychological assessment. All subjects underwent a follow-up review of approximately 24 months, and according to the diagnosis in the follow-up stage, MCI subjects were divided into converters to AD (MCI-c, 𝑛 = 21) and nonconverters (MCInc, 𝑛 = 10). 2.2. Image Acquisition. All participants were scanned within a single session on a 3.0T Trio Siemens scanner at XuanWu Hospital, Capital Medical University. Resting-state functional images were collected using an echo-planar imaging sequence with the following parameters: repetition time (TR) = 2000 ms, echo time (TE) = 40 ms, flip angle (FA) = 90∘ , number of slices = 28, slice thickness = 4 mm, gap = 1 mm, voxel size = 4 × 4 × 4 mm3 , and matrix = 64 × 64. Participants were asked to lie quietly in the scanner with their eyes closed during data acquisition. Each scan lasted for 478 s. For registration purposes, high-resolution anatomical images were acquired using a 3D magnetization-prepared rapid gradient echo (MPRAGE) T1-weighted sequence with the following parameters: TR = 1900 ms, TE = 2.2 ms, inversion time (TI) = 900 ms, FA = 9∘ , number of slices = 176, slice thickness = 1 mm, voxel size = 1 × 1 × 1 mm3 , and matrix = 256 × 256. 2.3. Data Analysis 2.3.1. Image Preprocessing. Image preprocessing was performed by using SPM8 (http://www.fil.ion.ucl.ac.uk/spm/) and Data Processing Assistant for R-fMRI [24]. The preprocessing procedures were performed including removal of the first 10 volumes, slice timing, and head motion correction. All data used in this study satisfied the criteria of spatial movement in any direction < 3 mm or 3∘ and the subjects demonstrated no

Neural Plasticity significant group differences in the head motion parameters (i.e., three translation and three rotation parameters). To normalize the fMRI data spatially, the T1-weighted images were firstly registered to the mean functional data, and the resulting aligned T1 data set was segmented and transformed into MNI space using the DARTEL toolbox [25] and a group template was generated. Next, the motion-corrected functional volumes were specially normalized to the group template using the transfer parameter estimated by DARTEL segmentation and resampled to 3 mm isotropic voxels. Further, the functional images were spatially smoothed with a 4 mm Gaussian kernel. The linear detrend and temporal bandpass filtering (0.01–0.08 Hz) was performed to reduce the influences of low-frequency drift and high-frequency physiological noise. Finally, several nuisance signals were regressed out from the data, including the six motion parameters, the global, the white matter, and the cerebrospinal fluid signals. 2.3.2. Whole-Brain Functional Connectivity Strength. To perform the whole-brain RSFC analysis, Pearson’s correlations between the time courses of any pairs of voxels were first computed, resulting in a whole-brain connectivity matrix for each participant. This procedure was limited within a gray matter (GM) mask, which was generated by thresholding (cutoff = 0.2) the mean map of all GM maps involving all subjects without cerebellum. These individual correlation matrices were then transformed as a 𝑧-score matrix by using Fisher’s 𝑟-to-𝑧 transformation to improve normality. We computed the FCS as the sum of the connections between a given voxel and all other GM voxels. This computation was conservatively restricted to connections with a correlation coefficient above 0.2, which could eliminate the weak correlations possibly arising from noise. 2.3.3. Seed-Based Functional Connectivity. To examine the detailed RSFC differences between MCI-c and MCI-nc, we performed seed-based connectivity analyses, using the clusters showing significant between-group difference on FCS as the seeds (i.e., left angular gyrus and middle occipital gyrus). Briefly, the mean time course within each seed was extracted by averaging the time courses of all the voxels belonging to the seed. Subsequently, the mean time course was further used to compute correlation coefficients with the time courses of all GM voxels. Notably, the computation was constrained within a custom GM mask that was made by thresholding (a probability threshold of 0.2) the GM probability map obtained in DARTEL segmentation. The resulting correlation coefficients were then converted to 𝑧-scores using Fisher’s 𝑟to-𝑧 transform to improve normality. For each MCI patient, we obtained two 𝑧-score maps indicative of the intrinsic RSFC patterns of the two seeds (i.e., left angular gyrus and middle occipital gyrus) based on the previous results of the group difference on FCS. Notably, given the ambiguous biological interpretations of negative functional connections, the statistical analysis for RSFC was restricted to positive connections. 2.3.4. Statistical Analysis. A one-way analysis of covariance (ANCOVA) was performed to determine the main effect of

3 groups on FCS, with age and gender as covariates, followed by two-sample 𝑡-tests post hoc analyses. The result for ANCOVA was thresholded at 𝑃 < 0.05 with a cluster size of 1350 mm3 , corresponding to a corrected 𝑃 < 0.05. The two-sample 𝑡tests post hoc analyses were performed within the regions showing significant group effects, and the threshold was set at 𝑃 < 0.05 with a cluster size of 324 mm3 , corresponding to a corrected 𝑃 < 0.05. Furthermore, in the AD pathologyrelated group, to determine the difference between MCIc and MCI-nc, we performed a two-sample 𝑡-test on FCS maps of the MCI-c and MCI-nc within the regions showing significant differences of AD against controls. The significant level was set at 𝑃 < 0.05 with cluster size of 216 mm3 , corresponding to a corrected 𝑃 < 0.05. All the cluster sizes were determined by Monte Carlo simulations [26] using the REST AlphaSim utility [27]. The two-sample 𝑡-tests were performed on the RSFC maps for each seed, with age and gender as covariates. The significant level was set at 𝑃 < 0.05 with a cluster size of 1350 mm3 , corresponding to a corrected 𝑃 < 0.05. The analysis mask was generated by selecting the voxels that showed significant positive RSFC in any of the two groups. To investigate the relationship between FCS and cognitive behavior, we performed general linear model analysis (dependent variable: FCS; independent variable: clinical variables, including MMSE, MoCA, AVLT-immediate recall, AVLT-delayed recall, and AVLT-delayed recognition) in the combined AD and MCI group with age and gender treated as covariates within the regions showing group effect. The statistical threshold was set to 𝑃 < 0.05 with a cluster size of 324 mm3 , which corresponded to a corrected 𝑃 < 0.05. 2.3.5. Discriminate Analysis. To assess whether the discovered differences of FCS and RSFC between MCI-c and MCInc could serve as the features to identify MCI-c patients from MCI-nc patients, we used support vector machine (SVM) as classifier to distinguish patients of the two groups. The features were selected as the values of voxels showing significant between-group differences, including the FCS and the wholebrain functional connectivity of the left angular gyrus and middle occipital gyrus. The leave-one-out cross-validation (LOOCV) was then used to estimate the performance of our classifier. In LOOCV, each sample was designated as the test sample, while the remaining samples were used to train the classifier. Accuracy, sensitivity, and specificity can be defined on the basis of prediction results of LOOCV to quantify the performance of the classifier: accuracy =

TP + TN , TP + FN + TN + FP

sensitivity =

TP , TP + FN

specificity =

TN , TN + FP

(1)

where TP, FN, TN, and FP denoted the number of MCI-c patients correctly predicted, the number of MCI-c patients

4

Neural Plasticity Table 1: Demographics and clinical characteristics of the participants.

Age (years) Gender (M/F) Education years MMSE MoCAc CDTd AVLT-I AVLT-D AVLT-R

AD (𝑛 = 25) 51–88 (69.4 ± 11.1) 9/16 0–17 (8.3 ± 5.4) 6–24 (16.8 ± 4.7) 5–22 (12.8 ± 4.8) 0–3 (1.7 ± 1.1) 0–5.7 (3.6 ± 1.5) 0–4 (0.6 ± 1.1) −2–8 (3.4 ± 3.1)

MCI (𝑛 = 31) 50–82 (67.9 ± 9.5) 14/17 0–21 (10.1 ± 5) 17–29 (23.5 ± 2.9) 9–24 (18.3 ± 3.9) 0–3 (1.8 ± 0.8) 2–7 (4.6 ± 1.3) 0–7 (2.7 ± 2.2) −3–13 (7.1 ± 3.9)

Control (𝑛 = 42) 51–79 (65.6 ± 7.1) 15/27 0–18 (11.1 ± 4.9) 20–30 (28.0 ± 2.3) 19–30 (26.0 ± 2.8) 1–3 (2.9 ± 0.4) 6–14.7 (9.3 ± 2.1) 4–15 (10.4 ± 3.0) 7–15 (12.4 ± 2.1)

𝐹 or 𝜒2 value 𝐹(2,95) = 1.52 2 𝜒(2) = 1.52 𝐹(2,95) = 2.4 𝐹(2,95) = 93.04 𝐹(2,73) = 81.32 𝐹(2,87) = 23.39 𝐹(2,95) = 108.87 𝐹(2,95) = 159.79 𝐹(2,95) = 72.48

𝑃 value 0.22a 0.67b 0.10a

Abnormal Resting-State Functional Connectivity Strength in Mild Cognitive Impairment and Its Conversion to Alzheimer's Disease.

Individuals diagnosed with mild cognitive impairment (MCI) are at high risk of transition to Alzheimer's disease (AD). However, little is known about ...
2MB Sizes 1 Downloads 16 Views